Abstract: A method includes receiving an input audio signal that corresponds to utterances spoken by multiple speakers. The method also includes processing the input audio to generate a transcription of the utterances and a sequence of speaker turn tokens each indicating a location of a respective speaker turn. The method also includes segmenting the input audio signal into a plurality of speaker segments based on the sequence of speaker tokens. The method also includes extracting a speaker-discriminative embedding from each speaker segment and performing spectral clustering on the speaker-discriminative embeddings to cluster the plurality of speaker segments into k classes. The method also includes assigning a respective speaker label to each speaker segment clustered into the respective class that is different than the respective speaker label assigned to the speaker segments clustered into each other class of the k classes.

Abstract: Embodiments described herein are directed to securing network-based compute resources. The foregoing may be achieved by determining a tag representative of non-malicious network addresses. The tag is determined by analyzing network data traffic received by a plurality of compute resources. Machine-learning based techniques may be used to automatically classify each network address that communicates with a particular compute resource as being malicious or non-malicious. Determined non-malicious network addresses for a particular compute resource are automatically associated with a tag. The tag is used to configure a firewall application to prevent access to a corresponding compute resource by malicious network addresses not represented by the tag. The number of non-malicious network addresses associated with a tag may be expanded by clustering compute resources having a similar set of network addresses that communicate therewith.

Abstract: Embodiments described herein are directed to securing network-based compute resources. The foregoing may be achieved by determining a tag representative of non-malicious network addresses. The tag is determined by analyzing network data traffic received by a plurality of compute resources. Machine-learning based techniques may be used to automatically classify each network address that communicates with a particular compute resource as being malicious or non-malicious. Determined non-malicious network addresses for a particular compute resource are automatically associated with a tag. The tag is used to configure a firewall application to prevent access to a corresponding compute resource by malicious network addresses not represented by the tag. The number of non-malicious network addresses associated with a tag may be expanded by clustering compute resources having a similar set of network addresses that communicate therewith.

Abstract: A system for automated rendezvous, docking, and capture of autonomous underwater vehicles at the conclusion of a mission comprising of comprised of a docking rod having lighted, pulsating (in both frequency and light intensity) series of LED light strips thereon, with the LEDs at a known spacing, and the autonomous underwater vehicle specially designed to detect and capture the docking rod and then be lifted structurally by a spherical end strop about which the vehicle can be pivoted and hoisted up (e.g., onto a ship). The method of recovery allows for very routine and reliable automated recovery of an unmanned underwater asset.

Abstract: A system for automated rendezvous, docking, and capture of autonomous underwater vehicles at the conclusion of a mission comprising of comprised of a docking rod having lighted, pulsating (in both frequency and light intensity) series of LED light strips thereon, with the LEDs at a known spacing, and the autonomous underwater vehicle specially designed to detect and capture the docking rod and then be lifted structurally by a spherical end strop about which the vehicle can be pivoted and hoisted up (e.g., onto a ship). The method of recovery allows for very routine and reliable automated recovery of an unmanned underwater asset.

Abstract: A system for automated rendezvous, docking, and capture of autonomous underwater vehicles at the conclusion of a mission comprising of comprised of a docking rod having lighted, pulsating (in both frequency and light intensity) series of LED light strips thereon, with the LEDs at a known spacing, and the autonomous underwater vehicle specially designed to detect and capture the docking rod and then be lifted structurally by a spherical end strop about which the vehicle can be pivoted and hoisted up (e.g., onto a ship). The method of recovery allows for very routine and reliable automated recovery of an unmanned underwater asset.

Abstract: A system for automated rendezvous, docking, and capture of autonomous underwater vehicles at the conclusion of a mission comprising of comprised of a docking rod having lighted, pulsating (in both frequency and light intensity) series of LED light strips thereon, with the LEDs at a known spacing, and the autonomous underwater vehicle specially designed to detect and capture the docking rod and then be lifted structurally by a spherical end strop about which the vehicle can be pivoted and hoisted up (e.g., onto a ship). The method of recovery allows for very routine and reliable automated recovery of an unmanned underwater asset.

Abstract: Software for storing and distributing media in a local area network is disclosed. The software comprises library software and player. The library software manages encrypted media files, for example, video files, and sends them to the player on request. The media files may be chapterised and the player may request one or more chapters. The player decodes the encrypted file and displays it. The player improves performance by using predictive chapter buffering, requesting the transfer of the next chapter from the library at time so that the file is completely received and ready to play at the time the current media playing is complete. The player minimizes the number of requests to the library by automatically creating a public folder of received files so that requests for one of the received files in the folder are transferred from the library to the player.

Abstract: A template for positioning electrical outlet and switch boxes for attachment to a stud in a new construction. An elongated flat plate has a series of vertical notches formed therein on either side, one of each representing the standard height for a wall outlet box, utility outlet box, thermostat, and switch box. The template includes guide means to straddle the conventional 2 .times. 4 stud. Vertically positioned across the lower and upper portion of each notch is a cross plate secured thereto a distance adjustable to 1/8 inch, 3/8 inch, 1/2 inch, or 5/8 inch to act as a horizontal depth finder for securing the box.